Hepatic Mechanisms of Action in the Triglyceride-Lowering Effect of Omega-3 Fatty Acid-Based Therapies

The atherosclerotic cardiovascular disease (ASCVD) risk posed by hypertriglyceridemia has become clear with Mendelian randomization studies, which use genetic variants in the human population to naturally assign individuals to a “control” or “treatment” group. Single nucleotide polymorphisms in loci such as APOC3 and ANGPTL3 that lower plasma triglycerides (TG) are significantly associated with CVD risk.¹ Moreover, exome sequencing studies that examine protein coding variants in these genes have also indicated a protective effect of lowering triglycerides.² However, the promising therapies targeting apolipoprotein C-III and angiopoietin-like 3 are currently monoclonal antibodies or antisense oligonucleotides that are not only expensive, but need to be administered by injection.^3-5 Pharmacotherapies based on n-3 fatty acids (FA) present orally-available options that are relatively affordable. As detailed in the AHA Science Advisory by Skulas-Ray et al., n-3 FA-based therapies are also efficacious in lowering triglycerides in hypertriglyceridemic individuals.

The main mechanism of action for n-3 FA-based therapies is likely the inhibition of sterol regulatory element binding protein-1 (SREBP-1), the main transcriptional activator of de novo lipogenesis (DNL) (synthesis of new fatty acids). In humans with non-alcoholic fatty liver disease (NAFLD), which is considered the hepatic component of the metabolic syndrome, DNL is increased more than 3-fold,⁶ indicating excessive SREBP-1 activity. SREBPs are synthesized as a precursor form that is a polytopic membrane protein that interacts with SREBP cleavage-activating protein (SCAP) and is retained in the endoplasmic reticulum (ER) by another transmembrane protein, insulin-induced gene (INSIG). Upon insulin stimulation, SCAP escorts SREBP to the Golgi apparatus where it undergoes proteolytic cleavages to yield a soluble domain that translocates to the nucleus to direct lipogenic gene expression.⁷

Studies in cell culture and rodents have proposed multiple molecular mechanisms for how polyunsaturated fatty acids (PUFAs) of multiple species inactivate SREBP-1.^8-10 The ER-localized ubiquitin-like domain-containing protein d8 (Ubxd8) polymerizes in response to unsaturated fatty acids, which protects INSIG from ER-associated degradation.¹¹ The stabilization of INSIG helps to sequester SREBP-1 in the ER, where it cannot be processed to its transcriptionally active form. Such a mechanism is also expected to affect SREBP-2, which directs the synthesis of cholesterol and increases low density lipoprotein receptor (LDLR) expression. However, n-3 FAs have repeatedly been shown to be specific for SREBP-1 activation. One group reported that docosahexaenoic acid (DHA) activates ERK, which leads to the phosphorylation of nuclear SREBP-1. Phosphorylation accelerates the turnover of nuclear SREBP-1 by proteasomal degradation, thus limiting the activation of lipogenic genes. Interestingly, the protein stability of nuclear SREBP-1 is much more sensitive to DHA than eicosapentaenoic acid (EPA).¹² Both DHA and EPA have been documented to reduce Srebf1 mRNA,^10,12 possibly because SREBP-1 drives the expression of its own transcript.^8,13 Figure 1 pictorially depicts the mechanisms by which n-3 FA can inhibit SREBP-1 activation.

Figure 1. Inhibition of SREBP-1 activity by n-3 FA. The effects of n-3 FA are shown in red symbols. The precursor form of SREBP-1 (orange) is retained along with SCAP (green) in the ER by INSIG (purple). Unsaturated FA promotes Ubxd8 (yellow) polymerization in the ER, which prevents it from mediating the proteasomal degradation of INSIG. Additional mechanisms not yet elucidated by inhibit the transport of SREBP-1/SCAP from the ER to the Golgi apparatus. In the Golgi, SREBP-1 is processed by two proteases (blue), which releases the nuclear (N) domain that directs transcription in the nucleus. DHA promotes the proteasomal degradation of the nuclear form of SREBP-1. In addition to lipogenic genes, SREBP-1 promotes the expression of its own transcript. Therefore, by these various mechanisms that act to suppress the amount of nuclear SREBP-1, overall hepatic levels of SREBP-1 levels also decrease.

These findings in rodents appear to be translatable to humans. In subjects taking 4 g/day EPA + DHA, who had >2% DHA enrichment in their erythrocytes (a reliable surrogate of hepatic DHA enrichment), the fraction of DNL-derived TG in the very low density lipoprotein (VLDL) fraction was halved.¹⁴ However, despite the elevated DNL rates in patients with insulin resistance, exogenous FA derived from lipolysis of adipose stores or chylomicrons continue to comprise the majority of VLDL-TG.¹⁵ An important SREBP-1 transcriptional target is glycerol-3-phosphate acyltransferase 1 (GPAT1), which catalyzes the first committed step in hepatic TG synthesis using exogenous FA. In mice, DHA supplementation to suppress SREBP-1 activity decreased Gpat1 expression 3-fold and normalized hypertriglyceridemia.¹⁶

The reason for increased LDL-cholesterol with some n-3 FA formulations in hypertriglyceridemic individuals remains unclear. As alluded to earlier, n-3 fatty acids have never been demonstrated in vivo to affect the activation of SREBP-2, which drives the expression of LDLR and proprotein convertase subtilisin/kexin type 9 (PCSK9). PUFAs have been shown in vitro to bind the oxysterol sensor liver X receptor (LXR),¹⁷ which could possibly raise LDL-cholesterol by lowering ATP-binding cassette G5/8 (ABCG5/8) expression and therefore biliary cholesterol excretion, but this has not been shown to occur in vivo.¹⁸ DHA has been shown to be a ligand for the bile acid nuclear receptor farsenoid X receptor (FXR).¹⁹ The FXR agonist obeticholic acid has been reported to increase LDL-cholesterol in humans,²⁰ which contradicts the findings in rodent models. Mice also have a poor response to statins, which precludes any detailed mechanistic studies on the lipoprotein effects of n-3 FA in the setting of statin treatment, even though it would be more reflective of the patient population.

All of the above-described molecular mechanisms can be ascribed to both DHA and EPA, so it is puzzling that EPA-only formulations such as icosapentyl ethyl esters (IPE) seem to have a different effect on LDL-cholesterol. However, it may be premature to conclude that EPA-only formulations are better for avoiding increases in LDL-cholesterol. The MARINE, ANCHOR, and REDUCE-IT trials on IPE all used light liquid paraffin (mineral) oil as the placebo control. Mineral oil is known to block the absorption of many medications, including statins. Consistent with this effect, increases in the particle concentration of most apolipoprotein B-containing lipoproteins were observed in the placebo group of the MARINE trial.²¹ Interestingly, when the subjects were split up according to statin treatment, there was a trend for an enhanced LDL-lowering effect of IPE in those taking statins.²¹ While the statin-treated groups in MARINE were small, this observation is suggestive of an effect of compromised statin absorption in the placebo group. Tellingly, the open-label JELIS trial that did not use a mineral oil control reported no differences in LDL-C.²² Therefore, the results expressed as placebo-adjusted values may not be a true reflection of EPA-only formulations’ effect on LDL-C.

In addition to improving hypertriglyceridemia, treatment with n-3 FA-based therapies may lower ASCVD incidence by addressing another risk factor. A recent large meta-analysis suggests that NAFLD is an independent risk factor for CVD, even after adjusting for traditional risk factors including body mass index, smoking and LDL-cholesterol.²³ The WELCOME trial on Lovaza showed demonstrate that 4 g/day EPA + DHA was efficacious in reducing liver fat.²⁴ A sub-study of the WELCOME trial reported a direct relationship between change in liver fat percentage and carotid intima-media thickness (CIMT) progression. There was also a direction relationship between a marker of hepatic necroinflammation and CIMT progression. Both of these correlations were independent predictors of CIMT progression even after adjusting for plasma TG concentrations.²⁵ The use of n-3 FA-based therapies may therefore yield benefits in addition to plasma TG lowering in the patient who is at risk of ASCVD.